SVM results exploration
Emeline Favreau
2021/05/29
Copyright 2021 Emeline Favreau, University College London.
Objective of analysis
We tested each species againt a dataset comprising gene expression data for the five other species, or a lineage specific dataset (e.g. only bees). The SVM model gives the list of predictor genes (genes whose expression predicts the phenotype of a given sample), after Recursive Feature Selection (ie iteratively removing the weakest gene in predicting phenotype, running from 3718 orthogroups to 100).
Analysis steps:
- Obtaining data
- Aim 1-6: Plotting results from the leave-one-out analyses with a training set of 5 species
- Aim 7-9: Plotting results from the leave-one-out analyses with a training set of bees only
- Aim 10-12: Plotting results from the leave-one-out analyses with a training set of wasps only
Aim 1: Ceratina australensis trained on 5 species
The larger line is the average for a given phenotype: Reproductive (R) or Non-Reproductive (NR).
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 2: Ceratina calcarata trained on 5 species
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 3: Megalopta genalis trained on 5 species
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 4: Liostenogaster flavolineata trained on 5 species
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 5: Polistes canadensis trained on 5 species
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 6: Polistes dominula trained on 5 species
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 7: Ceratina australensis trained on bees
The training dataset comprises of only two species (the other two bees).
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 8: Ceratina calcarata trained on bees
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 9: Megalopta genalis trained on bees
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 10: Polistes canadensis trained on wasps
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 11: Polistes dominula trained on wasps
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in imageAim 12: Liostenogaster flavolineata trained on wasps
## `summarise()` has grouped output by 'phenotype'. You can override using the `.groups` argument.
## Saving 7 x 5 in image